COMBINED M E T H O D S IN I N D O L O G Y by D. D.

KOSAMBI
Poona

This note suggests that the linguistic study of problems of ancient Indian culture would be more fruitful if supplemented by intelligent use of archaeology, anthropology, sociology and a suitable historical perspective. 1 Available Indian data in each of the fields listed need to be augmented by a great deal of honest and competent field work. None of the various techniques can, by itself, lead to any valid conclusion about ancient India; combined operations are indispensable.
1. Preliminary: The main idea back of the suggestion is that people

who live alike tend often to act and to think alike, especially if their historical development has followed parallel courses. Indian peasants in villages far from any city live in a manner closer to the days when the Pur~n.as were written than do the descendants of the brahmins who wrote the Pur~n.as. A stage further back are the pitiful fragments of tribal groups, usually sunk to the level of marginal castes; they rely heavily upon food-gathering and have the corresponding mentality. The existence of such differences is ignored by the Indian intelligentsia, to the detriment of its reasoning. In the judgement of Louis de la ValI6e Poussin :2 "Les savants de l'Inde sont excellents pour la lecture des textes, l'&ude des dates, etc. Mais quelques-uns sont bien les neveux des philosophes bouddhistes ou brahmanisants. A ceux-ci toute explication est bonne d~s qu'elle est sp6cieuse, et ils jouent avec des abstractions du 1 The reader is referred to two works of mine, entitled; (a) Introduction to the study of lndian History (Bombay, 1956) and (b) Myth & Reality; studies in the formation of Indian culture (Bombay, 1962), where further references will be found. Facts about Mahgr~.s~rian villages or customs from my own observations in the field are not documented. L. de la Vall6e Poussin, Dynasties et Histoire de l'Inde depuis Kanishka jusqu'aux invasions musulmanes (Paris, 1935), avant-propos, p. xix, footnote. Special attention is called to pp. 360-361 of this book, for important additional remarks on the process of brahminization.

178

D . D . KOSAMBI

second degr6 comme avec des r6alit6s concr6tes". This criticism, unfortunately too true, applies not only to Indian savants. The brahmanising tendency has seriously affected many distinguished foreign scholars whose long and exclusive concentration upon brahmin documents seems to have impaired their ability to distinguish between myth and reality. One consequence of such neglect may be seen in the formulation of "Hindu" Law. This type of jurisprudence is mainly brahmin traditional usage on property rights and inheritance. The sm.rti injunction (Ms. 8.41) that judicial (dharma) decisions were to be given only after due consideration of the - particular law and the usage of the region, caste-group and family group, guild etc. was apparently followed for a long time? However, no written record exists of any cases tried under this heterogeneous system. No attempt was made even by the British to study and collate the various caste-laws carefully as a preliminary for Indian common law. New forms of property were regulated under the foreign (British bourgeois) law; crime by an arbitrary penal code. The caste sabhds continue to function off the record, with diminishing force and powers. When the question of Hindu widow remarriage was being violently argued by reformers at the beginning of this century, even the most scholarly (like R.G. Bh~m..d~rkar) looked only to correct interpretation of the sacred texts, from the .Rgveda down. That 85 ~o of the population in their immediate locality allowed widows to remarry (and permitted divorce when either party felt aggrieved) made no impression upon the scholars nor upon the authorities on Hindu Law. P. V. Kfm.e's monumental history 4 of the Dharmaidstra meticulously restricts the discussion to sm.rti documents, avoiding any disagreeable contact with anthropology, sociology, or reality. This tunnel vision persists in all disciplines concerned with Indology. Field work has one disadvantage for arm-chair linguists. The amazing deftness with which world-shaking conclusions can be drawn without moving out of the study becomes less serviceable. I was told by a good linguist that the rather unusual Mar~t.hi village name of G o m ~ i (gad-fly s The Krtyakalpataru of Bhatta Laksmidhara (a minister of king Govindacandra Gaha.dav~da of Kanauj); Gaekwar Oriental Series CXIX for vol. 12 of the work, being the vyavahara-kd.nd.a. No special praetor peregrinus existed, and no ius gentium seems ever to have been officiallyrecorded or codified, though its existencein practice is clear. P.V. Kh.ne, A History of Dharmagdstra (Ancient and mediaeval religious and civil law), 5 volumes (still incomplete), Poona, 1930-1962. Though the vast majority of India's people are ~fidras in this classification, there is no way to determine just what ~fidras were actually meant by the few authors who wrote on ~Qdra rites and legal usage.

COMBINED METHODS IN INDOLOGY

179

or cattle-fly) has its obvious etymology. The villagers, however, usually speak of the place as goarn, shortened from go-area. The actual spot so designated is a small cave near the village with a fine 6th century image of Buddha, also unique for the region. Gotama Buddha had become Gotama r.si for local brahmins and the villagers follow the Prakrit form goama (+isi). Gomfi~i can thus be traced step by step to gotama--r.si, though the derivation at one jump seems to contradict accepted rules. The village name P~sa.ne is pronounced in half a dozen different ways within a range of twenty miles. The last syllable can vary, as in peasant Marfi.thi, from .ha to .ng, while the sa becomes a cerebral ca or the dental ta, for reasons that could not be discovered. Learned theses on Mar~t.thi continue to be written as if such difference did not exist; as if the rustic speech of S~ttara district were not markedly different from that of the adjoining Kofika.n. In Goa it was possible in 1925 for a keen ear to emulate Bernard Shaw's Pygmalion-Higgins and to locate a person's origin within five miles merely by his or her speech, which also gives away the speaker's caste or religion, status, profession and educational accomplishments to an observer who knows the locality. This diversity raises a natural question about the language of Asokan edicts. The local varieties have been determined by philological analysis ;5 the text of the same edict is not absolutely identical in different localities. This caused T. W. Rhys Davids ~ to declare that: "The Buddha and his followers adopted ... the particular form of this common speech ... that was current in Avanti'. Does the Pfili canon represent the idiom actually in the Buddha's mouth, through a collection made from oral tradition some two centuries after his death? The Buddha's strict injunction to his disciples to preach in the languages of the common people is either ignored or taken to mean that the said languages differed by no more than the various versions of the same edict. The discovery of the Shar-iKuna 7 (Kandah~r) edict in Greek and Aramaic (without a Mfigadhi equivalent), a brief r~sum6 of the standard Asokan declarations, changes the picture. It is difficult to believe that Greek and Aramaic were then the two languages of Afghanistan, though they were undoubtedly the two major languages and scripts which would reach the great majority of literate people passing through Kandahar. Asokan Prakrit and
s E. Hultzseh, Corpus Inscriptionum Indicarum I (The inscriptions of Asoka), Oxford, 1925 gives the complete Asokan texts known to that date, and a linguistic analysis. The Cambridge History of India, vol. I, Ancient India, ed. E. J. Rapson (Cambridge, 1922), p. 187. 7 First published in full in Serie Orientale Roma (ISMEO) XXI, 1958: Un editto bilinguale Greco-Aramaleo di Asoka. The treatment in JA, 1958.1-48 seems preferable.

180

D . D . KOSAMBI

Brfihmi have to be given the same position in the greater part of India, a country where the language must then have changed from one small valley to the next as it does in Assam today. The decrees were promulgated by the emperor, but the rescripts circulated by his predominantly Magadhan secretariat. It is not plausible that spoken M~gadhi had then so little inner variation as the pillar and rock texts show. During a walk of twenty miles in Goa, "want to go" changes from jdftm,ka hoy6 to vacam,ka jdya, while another twenty miles in the same direction reduces it to vacakd; this is for peasants of the same caste and status who manage nevertheless to understand each other. Patafijali s gives local usage in spoken Sanskrit (not different languages) of his day: "goes" was gavati in Kamboja, hammati in Sur~t.s.tra, ram.hati in the east (the Gangetic regions), but gamati for "real Aryans". Yet Sanskrit then possessed the standardization of an extensive literature, the scriptures being committed to memory without alterations of a single syllable or accent. In both cases, the reported variation is much greater than for the official Prakrit of Asoka. The analysis of the latter cannot therefore be put upon the same footing as the comparison of early Greek epigraphs, say Ionian, Attic, Doric and Cretan linear B. These were issued by independent local authorities in a land where the profusion of written contracts and registers afforded a striking contrast with india - where the natives' honesty and truthfulness in the absence of written agreements astounded Greek observers. 9 The Prakrit spoken by different characters in the M.rcchakat.ika has been separated into varieties labelled with local names. But even the M.rcchaka.tika Ca.nd.alas use a Prakrit easily understood by the rest, while the Ca.n.d~das of the Jfitakas spoke a language among themselves incomprehensible to "Aryans". The parallel is with the idioms used by a Welsh or Irish character in a modern English play as against the actual Welsh language or Erse. Though the variation is decidedly less than one would expect from Patafijali, the use of Prakrit is more natural in this particular drama than in other Sanskrit plays. Here, the Sfttradhftra declaims in Sanskrit to the audience, but lapses into Prakrit with his own womenfolk; much as educated Goans who consider Portuguese of Marft.thi to be their real language speak Kofika .ni to women and servants. No other Sanskrit drama makes so great a concession to everyday life, s Commenting on the vartika: sarve degantare: p. 65 of Patafijali's Vyakaraga Mahfibha.syawith Kaiyat.a'sPradipaand N~ge~a's Uddyota,vol. I, Nir~ay-S~tgarPress, Bombay, 1938. 9 The original remark may have been by Megasthenes, and is seen in its most forceful version in Arrian's "No Indian is ever known to lie".

COMBINED METHODSIN INDOLOGY

181

just as none other deals with a historical in preference to a mythical episode. Literary Prakrit with all its varieties had become standardized, five centuries after Asoka. The presumption is strong that the observed variation in Asokan Prakrit is due to clerks and officials of the secretariat rather than to common local usage; very few of the original inhabitants of Maski in Mysore could have mastered the Magadhan tongue. In modern science, it has been recognized that the variation is a very important characteristic of the material, particularly when dealing with living organisms. Fundamental methods developed by R. A. Fisher 1~ and others for taking such variation into mathematical account have led to great advances in biology. But I have yet to see any recognition of the philosophical principle, let alone the use of delicate statistical tests, in Indology. Still worse, most of our field work is done by educated men who often miss significant features or impose their own views upon the observed. In particular, the world of the women with its secret rites exclusively the property of female members of the group and the inevitable archaisms that mark the speech of the women when trade and intercourse with strangers is a male prerogative - all these inevitably escape observation, especially when the ritual has not been written down and the language not standardized by formal education.

2. Ibhya. India is a country of long survivals. It is known that the Buddha's birthplace was the sacred grove of a Mother-goddess still worshipped at the spot under the same name after two and half millennia; but the gftkyas and Buddhism have vanished from the locality. Literate Mah~tr~,.s.trians use the word len.~ ( = layanam) for a monastic cave, originally excavated as a retreat, and referred to in S~ttav~thana inscriptions under essentially the same name. To the peasantry near Kftrle caves the natural term is veher (often pronounced vyahar), from the Buddhist vihdra, which the caves actually were for centuries. Surprisingly enough, the term changes at Karhg.d (the ancient Karahft.taka) where the (6th century A.D.) Buddhist caves are called vavri, an archaic Sanskrit word whose filtering down to the lowest stratum of the population can only be explained by the strength of the brahmins at KarMt.d. The peasant dialect about Karhft.d is otherwise not more influenced by Sanskrit than elsewhere in Mahftrft.st.ra. The caves were carved out by a class of people intimately connected with the brahmins of a great trade centre.
lo R.A. Fisher, StatisticalMethods For Research Workers;10th ed. (Edinburgh, 1948).

182

D . D . KOSAMBI

These survivals n a t u r a l l y lead to the view t h a t there has been no real change in I n d i a over the ages. A m o n g the m o r e stupid displays m a y be m e n t i o n e d A. A. F t i h r e r ' s p u b l i c a t i o n n o f a p h o t o g r a p h o f T h a r u tribesm e n near the B u d d h a ' s b i r t h p l a c e as m o d e r n Sftkyas, t h o u g h there was n o t h i n g whatever in the t r i b a l n a m e or legends to indicate the equivalence. F a H s i e n ' s account 1~ showed t h a t b y the 4th century A . D . , the Sftkyan capital was virtually deserted. By the time o f H s i i a n T s a n g in the early 7th century, a B u d d h i s t revival seems actually to have r e l o c a t e d K a p i l a v a s t u several miles a w a y f r o m its original site, 13 if the two travellers' accounts (so accurate in detail) are to be reconciled. H o w m a n y tribes (before the Tharus) w a n d e r e d over the S~kyanjanapada r e m a i n s u n k n o w n . This "timeless u n c h a n g i n g E a s t " t h e o r y m a y insidiously distort the entire m e a n i n g o f a d o c u m e n t a n d thus reduce the value o f o u r a l r e a d y m e a g r e source material. F o r e x a m p l e : R.gveda 1.65.7 describes the fire-god A g n i : ibhydn nd rdjc~ vdndny atti " A s a king the ibhyas, so eats he (Agni) up the forests". K. F. G e l d n e r ~ translates this as " W i e der K 6 n i g die Reichen frisst er die HSlzer a u f " . The f o o t n o t e to this gives an alternative: " O d e r : W i e ein K 6 n i g seine Vasallen". S~ya.na c o m m e n t i n g o n the same .rk gives ibhyd gatrava.h .... yad vd dhaninal.z; tan yathd dhanam apaharan rdjd hinasti tadvat. Thus, G e l d n e r has t a k e n the second o f Sgya.na's alternatives for a w o r d t h a t ix A . A . Filhrer, Buddha Sakyamuni's Birthplace in the Nepalese Tarai (Allahabad, 1897). 12 A Record of the Buddhist Countries by Fa-hsien; Peking 1957. The translation in S. Beal's collection: Ta-Tang-Si-Yu-Ki: Buddhist Records of the Western World; 2 vols. London, 1884, also contains this in the introductory portion. 18 The discrepancy seems to have been first considered in archaeological detail by Vincent Smith in the prefatory note (p. 10) to P. C. Mukerji's "A report on a tour of exploration of the Antiquities in the Tarai, Nepal, the region of Kapilavastu during February and March 1899" (Arch. Surv. Ind. No. XXVI, pt. i, Imperial Series; Calcutta 1901). Smith was capable of identifying Satna railway station in central India with the site of ancient Kau~ftmbi (JRAS, 1898. 511), but his discussion of the Kapilavastu problem seems reasonable. The strictures on pp. 3-4 of the same preface about Fiihrer's supposed excavations are by no means excessive. la K. F. Geldner, Der Rig-Veda aas dem Sanskrit ins Deutsche iibersetzt und mit einem laufenden Kommentar versehen. Cambridge, Mass. (Harvard Oriental Series vols. 33-35) 1951. For the Sanskrit text and S~ya.na's commentary, I have used the 4-volume edition of the ggveda-Sa~hita issued by the Vaidik Sar0shodhan MaoOal., Poona 1933-1946. There may have been a sort of feudalism at a stage not much later than the l~gveda, among the Hittites: E. Neufeld; The Hittite Laws translated into English and Hebrew with commentary (London, 1951), particularly laws 39-41 ; 46-56 for military service as condition of land tenure. But there is no evidence for comparable fixed land settlement in the .Rgveda, nor for a king ruling over many different tribes by the military strength of a few of his own tribal comrades, as with the Hittites.

COMBINED METHODS IN INDOLOGY

183

occurs just once in the whole of the .Rgveda. That this did not entirely satisfy seems clear from his note on RV. 9.57.3 ibho rdjeva suvratd.h. The footnote here reads: "Die Verbindung von ibha, ibhya mit rdjan (1.65.7; 4.4.1. und hier) ist fiir beide WSrter bedeutsam and harrt noch der sicheren LSsung. Andererseits ist die Bedeutung "Elefant" ftir ibha, "reich" fiJr ibhya durch das sp/itere Sanskrit (rdjd ibhena Manu 8.34!) so gesichert, dass sic kaum zu umgehen ist. ibhya wird sich zu ibha verhalten wie dhdnya zu dhdna. P~li ibbha in der bekannten Formel (s.P.D.) und ibha in Ch~nd. Up. 1.10.1-2 sind aus dem Zusammenhang nicht mehr sicher zu bestimmen ... Lehnt man aber die klassische Bedeutung ftir den Veda ab und sucht den Sinn in der von Roth gewiesenen Richtung, so empfiehlt sich statt "Gesinde, HSrige" (Roth) vielmehr fiir /bha und ibhya "Vasall". ibho rdjd w~ire dann der VasallenkSnig". This is a valiant attempt made by a scholar of merit to settle the meaning of a unique term in a document which he had studied intensively for so many years. The basic question is whether .Rgvedic society had kings who ruled absolutely over vassals and over elephant-owning noblemen. It would seem extremely unlikely, taking the hymns as a whole. On the other hand, if the meanings of ibhya could be more closely determined, a certain amount of history emerges from the verse in question. The matter could have been settled by Asoka's 5th Rock Edict which is clearly legible for the relevant portion at Dhauli, Shahbazgarhi, Kalsi and Mansehra. There, bambhanibhesu is beyond question an antithetic compound, like the preceding, "masters and servants". One should expect that the ibbha here would be the lowest of castes, as the brahmin was the highest. However, the point may still be argued, and Jules Bloch, 15 for example, deliberately leaves the word untranslated, as he does every other word that might contradict the idea that Asoka was a pious dotard bent upon preaching Buddhism. So, we might look closer at the two sources which seemed indecisive to Geldner. The Pall Dictionary ~6of Rhys Davids gives ibbha primarily as the lowest of menials, lowest of the low. The context of the third sutta of the D~ghanikdya ~7 (Ambat.thasutta) makes it certain that ibbha is used as a term of abuse, to indicate the contempt in which some local brahmins held the Sakyans as men of low lineage. This meaning fits all contexts cited, and
15 J. Bloch, Les Inscriptions d'Asoka (Paris, 1950), p. 104. 1~ The Pali Text Society's Pall-English Dictionary; London, 1921. 17 The Pall Text Society's edition of the Digha Nikdya, vol. I (London, 1890), p. 90; Buddhaghosa comments: ibbhd ti gahapatika (peasants), on p. 234 of the Suma.~gala Vilasini vol. I (London, PTS 1886).

184

D.D. KOSAMBI

is generally accepted. The only other meaning given by that dictionary is late, in a comment of Buddhaghosa on the Jdtakas. As for the Chandogya Upani.sad reference, there seems to me no doubt of the meaning of ibhya in its particular context. The story is of a brahmin Ugasti Cakrftyan.a of the Kuru country, who was wiped out by a plague of locusts (mat.acf-hata; commentators prefer "hailstorm"). At a village of ibhyas, he saw an ibhya eating kulmds.a broth, begged the leavings (which his wife could not bring herself to eat, famished as she was) and from the strength gained from this distressing meal, made a success the next day at the royal sacrifice. The commentary that passes under the name of Sam.kara gives for ibhya the alternatives "rich man" or "elephant-driver (of low caste)"; whereof Hume in his English translation takes the first. Gop~l~nanda-svftmi is in his comment gives only hastipak = elephantdriver for ibhya. Geldner may seem to appear justified in his assertion of ambiguity. But what is kulmd.sa? Neither lexica nor commentators make of this anything but food of the lowest grade. Whether my personal interpretation of kulmd.sa as the lowly vetch Glycine tomentosa is accepted or not, it was certainly not food for a nobleman rich enough to own elephants. The story has a point only if it shows the desperate straits to which a learned brahmin had been reduced. Not for the first time in our records, for Vamadeva in RV. 4.18.13 claims to have cooked a dog's entrails in hunger: dvartyd g~ma dntrd.ni pece. This .rk is put into Indra's mouth by Geldner, who here ignores the logically consistent brahmin tradition reported by S~tya.ha and by the Manusm.rti (10.106)to the effect that the degradation was Vamadeva's. Finally, what can a village of ibhyas (where an ibhya could be seen eating outdoors) mean, if not some hamlet inhabited by people of a low caste-guild? Such villages still exist. If you take ibhya as the equivalent of the tribal caste M~tafiga, the modern m~hg, originating from people with an elephant totem, every one of the passages discussed makes sense. The Aryan king of RV.I.65.7 would eat up tribal savages mercilessly. The brahmin could take soiled food from the lowest caste only in times of unutterable famine.

3. Sdmanta. Naturally, this raises the question of feudalism in India: When did vassals and feudal barons as such come into existence? The
is The Nirtaayasfigar editions of the Upani.sads have been used for the text. With the commentary of Gopfilgnanda-sv~tmi, NSP 1932. For the English translation, R. H. Hume, Tire thirteenprincipal Upanishads translatedfrom the Sanskrit (Oxford, 1934).

COMBINED METHODS IN INDOLOGY

185

Sanskrit word to be discussed is the post-vedic sdmanta, meaning originally "neighbour" or "neighbouring ruler". In his indispensable translation of the Artha~dstra 1~, J, J. Meyer generally takes this in its later meaning 'vassal'. If the translation is justified, then India was unique in having a feudal system about a thousand years before Europe, or the document is a late forgery. But no one puts the book later ~~than 300 A.D., and the question must be asked whether feudal barons were in existence even at that period. The J~ttakas show sdmanta only as "neighbour"; the feudal institution is absent. The few k.satrapas and maMk.satrapas known in inscriptions are actually or virtually independent kings. Fortunately, it is possible to date, within limits unusually narrow for India, the period when sdmanta acquired the meaning "feudal baron". We may note that even in the Arthagdstra, the word sdmanta has often the meaning "neighbour", without alternative - as for example in Arth. 3.9 when transfer of title to houses and plots of land is in question. However, in every single case, sdmanta can consistently be translated as neighbour, whether royal or commoner, without incompatibility, in fact, in Arth. 6.1., Meyer contradicts himself by translating gakyasdmanta.h at the beginning as "Herr fiber seine Vasallen" and in the middle of the same chapter as "yon Grenznachbarn umgeben, die man in der Gewalt hat". The latter translation would fit both contexts, the former would not. There is no sdmanta baron in the Manusm.rti. The earlier Guptas rule over no sdmantas in their inscriptions; the posthumous Haris.e.na pragasti 21 of Samudragupta on the Allahabed pillar mentions no barons. Dharasena of Valabhi who appears as the first mahdsdmanta ~ in A.D. 527 is an independent king friendly to the Guptas (from the tone of his inscriptions), not a peer of the realm. The Mandasor pillar ~8
19 J . J . Meyer, Das altindische Buch vom Welt- und Staatsleben; Das Artha~astra des Kaut.ilya (Leipzig, 1926); the text used has been the revised southern edition Kaut.aliyartha~astram (Mysore, 1960). 20 A. Berriedale Keith, A History of Sanskrit Literature (Oxford 1928), p. 461. The discussion in my history book (note i) and in JAOS. 78.169-173 may be referred to for the authenticity of the Artha~astra. 21 J. F. Fleet, Inscriptions of the early Gupta Kings and their successors; Corpus Inscriptionum Indicarum Ill(Calcutta, 1888). The Haris.en.a pra~asti of Samudragupta (posthumous) is on pp. 6-17. 32 ibid. p. 41 of the introduction. As the founder of the Maitraka line was Bhat.arka, a senapati, the interpretation of mahasamanta as 'duke' would have been justified only if the Valabhi kings made any reference to some Gupta emperor as suzerain. ~a ibid. inscription No. 33, pp. 146-8. Line 5 of the inscription has samantair yasya bahu-dravi.na-h.rta-madaiO padayor anamadbhis, but these samantas are explicitly mentioned as coming fl'om territories over most of which Ya~odhannan had set up no administration and could claim no permanent sovereignty, namely from the Himalaya

186

D . D . KOSAMBI

inscriptions of Ya~odharman, who drove Mihiragula and the Hurts out of M~lw~t, say that the king defeated and humbled all the sdmantas, which can only mean neighbour kings. But the Vi.s.nu.se.na charter 24 of 592 A.D. takes sdmanta only in the sense of petty feudal viscounts who might press labour for corvre, or infringe upon the rights and immunities of merchants to whom the charter was granted. Thus, the change in meaning falls within a period of less than 60 years, say the second half of the 6th century A.D. It is confirmed by the Ten Princes ~5 of Dan.d.in, where sdmanta can only mean feudal baron, though the author shows remarkably close reading of the Artha~astra as of many other works. The copper plates 2e of Har.sa, supported by Chinese travellers' accounts, prove that feudal relationships and sdmanta " b a r o n " had come to stay. The entire structure of the A r t h a ~ s t r a , considered as a whole, contradicts the possibility of feudalism. The state collected its taxes in kind, but processed and made into commodities an enormous number of natural products thus gathered. The whole economy and the system of administration was based upon cash valuation, as may be seen by the minutely detailed table of fines and of salaries. Moreover, the state itself owned most of the land under the title of s~td, the rd.st.ra being still under private enterprise of various sorts though subject to imperial taxes. Neither in the mechanism of collecting taxes, nor in the administration of law and order, nor in military service is the sdmanta feudal officer mentioned; the respective officials are named, and have fixed monthly salaries paid in cash. The high ministerial mantrin and amdtya are also salaried posts not based upon hereditary tenure or nobility of rank. A "vassal" in the feudal sense would make the whole document logically inconsistent. As for the neighbouring rulers, the whole purpose of the Arthafftstra is to make its

to the oceans. The reference can only be to kings defeated in some passing raid or invasion, and this is strengthened by special mention in the preceding line of Hurts (Mihiragula) and other kings, whom even the Guptas could not vanquish but who were beaten by Ya~odharman. 34 D.D. Kosambi, Indian feudal trade charters, JESHO 2, 1959, 281-293. 25 Daga-kumara-caritam ofDa.nglin, 14th ed. by N. K. Go~tbole, Nir.nayasftgar Press (Bombay 1940), p. 184; but the remarkably silly comment samanta.h syad adhigvarab made by lifting half a phrase out of the Amarako~a confuses the issue. The context here, as in chapter 8 (p. 267ff.) does not leave the meaning in doubt. Keith (Hist. Skt. Lit. p. 297) suggests a date slightly before Har.savardhana for Dao0in, so towards the end of the 6th or beginning of the 7th century A.D. ~8 Epigraphia lndica 4.208-211, where mahasdmantas are named for the execution of the Banskhera plates of Har.sa. See also p. 130 of B~na's Har.sacaritam (7th ed. NSP Bombay, 1946), where only 'baron' will suit; on p. 100, 150 &c., mahasamantas of the court are named again.

COMBINED METHODS IN INDOLOGY

187

king the universal monarch, starting on level terms with the sdmantas. But conquest did not mean reduction of the beaten king to vassalage; he and his officials were to be maintained in their old position. No special tribute is mentioned. The profit of aggression came to the conqueror from the development of waste land as new s~t~iplus absolute control of mineral resources as a state monopoly. The land visualised is one divided into janapada territories, each originally belonging to a particular tribe, say Magadha, Kosala, Videha & c. These were separated by extensive forests infested by predatory tit.avika savages who were still in the food-gathering stage, difficult to conquer by military methods, or at least to conquer with due profit. In the intermediate stage were a few powerful, armed, tribal oligarchies. These had to be broken ruthlessly by every method at the king's command. There was no need or place for feudalism in any recognizable meaning of the word, in this type of state. Not only do these considerations furnish important data for Indian history, but they also help clarify points that remain unexplained or have escaped attention. The Allahabad pra~asti of Samudragupta21 says that he had reduced all forest kings to servitude: paricdrik~-k.rta- sarvdt.avika-rdjasya, and the context shows that this refers to Ary~varta, the Gangetic basin, probably including West Bengal. This finished the course of settlement begun by Magadhan kings before the Artha~stra, and accounts for the new prosperity of the Gupta empire. The great forest still existed in places, e.g. between Allahabad and Ban~ras, but had been cleared of armed savages; its reduction to farmland was a matter of time, no longer of armed intervention. Gupta gold coinage, beautiful as it is, supplements Chinese pilgrims' accounts to show that barter economy was becoming prevalent; Har.sa's coins are so few that the economic trend seems to have been virtually complete by the 7th century A.D. Other steps to feudalism were payment of officials by the income of specially assigned plots of land - impossible in the .Rgvedic economy (when fixed plots did not exist) and frowned upon by the Artha~stra. The definition of the paramount ruler: rdj~ tu pran.atd~e.sa-s~manta.h sydd adh~&ara.h in Amarako~a 2.8.2. fits only the Ya~odharman type of conqueror of neighbouring kings; sdmanta as "feudal baron" would not explain the given hierarchical order: adh~vara, cakravartin, sdrvabhauma; but if none of these, then a ma.n.dale~vara. It follows that the Amarako~a cannot be later 2~than the first half of the 6th century A.D. The tradition that places its writer at the same court as K~lid~sa seems quite reasonable so that
37 Keith (Hist. Skt. Lit. p. 413) vaguely places him about 700 A.D., but without committing himself.

188

I). D. KOSAMBI

the work may be as early as the late 4th century A.D. On the other hand, I had placed the poet Bhart.rhari in the opening centuries of the Christian era, which can be disproved on our deductions about the meaning of sdmanta. The Bhart.rhari stanza ~8 that begins bhrata.h ka.st.am aho (or sd ramyd nagarf in the southern recension) takes sdmanta as the high noble of a royal court, and is attested by all complete M S S . Inasmuch as the manuscript evidence also compels inclusion of the stanza bhavanti namrds tarava.h phalodgamai.h which is to be found in the S~kuntalam ~9 of Kalid~tsa (whereof the critical study needs to be extended), it follows that even the nucleus of the Bhart.rhari collection contains verses composed two centuries or more apart; the archetype restored on present M S evidence still remains an anthology. To round out the discussion, it can be shown that the transition from the .Rgvedic to the Artha~tstra society as we have reconstructed it was natural. The relevant documents are the various brdhman, as, from whose diffuse liturgical contents a useful collection of data has been boiled down by W. Rau. 3~ The king of this intermediate period was a small princeling, without very rich elephant-owning ibhya vassals. As the first among equals, he could be deposed. The move towards absolute rule unrestricted by tribal law was also evident. The ostracized (aparuddha) king appears again to intrigue in a somewhat more ambitious r61e in the Arthagdstra. Production on the land was, in each locality, in the hands of people with bonds of kinship, sajdta; this was the only form of association permitted on the Artha~stra ruler's s~td crown lands, and the text has been emended to sujdta (high-born, upper-caste) by heedless editors. The correct reading is confirmed by the fact that even under the Mughals, villages were generally tilled by a birddar~ (kinship group), and undisturbed villages (e.g. in Mah~tr~.s.tra) are still populated by people with the same clan-name, usually reminiscent of some totem (e.g. Magar, Lfm..dage, V~tji, More). While better developed than in the RV, the Yajurveda-Br~thma.na grdma was still a mobile association of human beings, who moved seasonally with their cattle to and from one territory to the other; very different indeed from the fixed agricultural village of today. The meeting of two such groups on the transhumance march
28 The Epigrams attributed to Bhartrhari (Singhi Jain Series No. 23, Bombay 1948) is the critical edition where the stanza may be seen as No. 169. ~9 ibid. stanza 63. In the 12th Nir.nays~tgar edition of the Abhij~ana~akuntalam (Bombay 1948, ed. N.R. Acarya), act 5, stanza 12, in Pisehel's Harvard Or. Ser. edition, 5.13. 80 WilhelmRau, Staat und Gesellschaft im alten lndien (Wiesbaden, 1957),particularly pp. 51-54.

COMBINED METHODS IN INDOLOGY

189

meant conflict, as the word sam.grdma for battle proves. If, now, we take Geldner's meaning for ibhya and Meyer's for sdmanta, the .Rgveda, the Brghman.as and the A r t h a ~ s t r a fail to give a consistent picture of developing Indian society.

4. Udumbara. Jean Przyluski (JA. 208. 1926. 1-59) describing the

Udumbaras as an ancient people of the Punjab, reached the conclusion: " O n peut donc admettre que Udumbara, Odumbara, Kodumbara sont les variantes d'un m~me nom d6signant un peuple austro-asiatique du Nord de l'Inde". The basic theory, again in Przyluski's words, seems to be as follows: "La r6partition des populations de l'Inde avant Alexandre aurait ~t6 le r~sultat de trois invasions successives. D'abord les Austroasiates recouvrent en partie l'616ment dravidien et ne laissent gu6re 6merger que l'ilot brahui au Nord et les masses du Dekhan au Sud. Puis les Aryens, descendus dans l'Inde par le Nord-Ouest, s'6tablissent progressivement dans les vall6es moyennes de l'Indus, de la Yamung, du Gange, et rayonnent autour de ces foyers de culture brahmanique. Plus tard enfin, les B~thlika, venus de l'Iran oriental, s'infiltrent, marchands et aventuriers, ches les tribus austroasiatiques laiss6es ~t l'6cart par les Brahmanes; en organisant de vastes conf6d6rations comme celle des gglva et en faisant circuler de l'Ouest/t l'Est leurs caravanes, ils pr@arent la formation des futurs empires et assurent la liaison de l'Inde et de l'Occident". These conclusions have caught on very well with a certain class of brahminising disciples, lovers of the "explication sp6cieuse" and "logique imperturbable". The Austro-asiatics are even credited 31 with the Indus valley civilization and that of Sumer! Rather than p.lunge into the linguistic morass, it might be more profitable to analyse the technical details of the three supposed pre-Alexandrian invasions. The British "invasion" of india reached maturity in approximately two
Suniti Kumfir Chatterji in The Bh~ratiya Itihftsa Samiti's History and Culture of the lndian people, vol. I 9 The Vedic Age, chapter VIII, for the statement o f the austrosl

asiatic hypothesis. On page 153: "We may admit the possibility of Sumerian and Austric being related, for we have to remember that the Proto-Australoids, who are supposed to have been the original speakers of Austric, were a very ancient offshoot of the Mediterranean race, and as such in their trek to India where they became specially characterized they may have left some of their tribes on the way, or some of their kinsmen might earlier have preceded them and had established themselves in Mesopotamia, to become the Sumerians who built up the basic culture of that part of the world. But even then it seems that India was the centre from which the Austric speech spread into the islands of the east and the Pacific; and the theory that there is actually an Austric Family of Languages in its two groups of Austronesian and AustroAsiatic, as propounded by Pater W. Schmidt, may be said to hold the ground still". I can't even understand this, let alone admit it.

190

D . D . KOSAMB1

centuries. Its ultimate cultural dominance and military success rested upon superior technique of production and a social form (the bourgeois) decidedly more efficient than feudalism. The Muslim invasion took six centuries to span comparable stages. The military technique is again well known while their developed feudalism was more efficient than the priestridden Indian system before them. In both cases, the success was out of all proportion to the actual number of invaders. There was no question of "submerging" the indigenous population, no matter how much Islam grew by conversion. So, Przyluski's three invasions prior to Alexander's ephemeral raid must have been much more powerful in numbers, not to speak of superiority in productive technique, military organization, and social form, relative to whatever existed in India at the time of each. The case for the Aryans supports these contentions at first sight. The older view that an "Aryan tribe" or "race" is as ridiculous a combination of attribute and noun as a "brachycephalic grammar" need hardly be considered. Strabo talks of Aryans on the banks of the Indus in Alexander's day; Darius 1 claims in his grave inscription to be an Aryan of Aryan descent: ariya, ariyacifa. So we need hardly go into the etymology of Hariana and Iran or speculate about the Germanic Arii in Tacitus. Archaeologists tell us that Aryan technique 32 as such does not mean any special type of pottery or tool; they picked up whatever suited them while smashing through the barriers of little atrophied peasant communities in Asia Minor. The military success of the first wave, dated 33 at about 1750 B.C., may be ascribed to the fast horse-chariot and a mobile food supply of good cattle. The second main wave at about the end 84 of the 2nd millennium B.C. added thereto the knowledge of iron, the first cheap 8~ V. Gordon Childe, The Aryans (London, 1926). The work needs revision, but the basic idea seems uncontradicted by new finds. 38 A.L. Oppenheim, The seafaring Merchants of Ur, JAOS 74, 1954. 6-17, a review analysis of Vol. V of the texts from L. WooUey'sexcavations at Ur, by H. H. Figulla and W. J. Martin: Letters and Documents of the Old-Babylonian period (London 1953). The break (due to an Aryan invasion) came about 1750B.C. if Melubba be the Indus valley; though so competent a scholar as S. N. Kramer would take Tilmfm as Harappg, it seems clear that the usual identification with Bahrein must stand. (JAOS. 74. 1954. 179). W. Wast, curiously enough, also placed the Aryan invasion of the Indus region as at about 1750B.C. (WZKM 34. 1927, p. 190), but this is simply a guess from poor archaeological material, without a scientific method for estimating the time from linguistic sources alone. a~ The two-wavetheory was confirmedby personal discussionwith Prof. S. P. Tolstoy, in 1955. The mention of I.s~g~va(=Vi~taspa?), I.s~aragmi, and Sugravas (Husravah) in the RV seemed to me philological evidence for the second wave; the archaeological basis in India may be the two layers of the Harappan cemetery H. Prof. Tolstov also showed Indian type of faces in Kushan frescos (note 39 below), and in a skull reconstruction.

COMBINED METHODS IN INDOLOGY

191

metal that made the heavy plough and extensive agriculture possible. This last point, of no importance to linguistic scholars, must be properly understood. In six African animal preserves, 85 the annual "production" of meat ranges from two tons to 34 tons per square kilometre. First class range land in Oklahoma yields 14 tons of beef per square kilometre annually; good Belgian meadowland runs to 45 tons. All this is with modern conservation and fire-arms. If the meat were to be procured by traps, pitfalls or bow and arrow, the actual yield would be much less; supplementing primitive weapons by bush fires would cause (and has elsewhere caused) great ecological changes which deplete the supply of game and therefore eventually the human population. Briefly, a change from hunting and food-gathering to a pastoral economy in suitable territory would support, say, eight times the population on the same land; plough farming could again multiply the number of people by at least as great a factor. Moreover, cattle-breeding and agriculture provide a regular food supply, where food-gathering is uncertain. Only the Indus region and part of the Gujar~tt loess area could have had any farming other than primitive slash-and-burn (Brandwirtschaft) or digging-stick cultivation before iron became plentiful. The river flowing through an alluvial desert in a tropical climate is of the utmost importance. That is why we find the first civilizations in Mesopotamia, on the Nile, the Indus; not on the Amazon nor the Mississipi. Next best would be a loess corridor, as in China and on the Danube. This explains why the Ganges and Yamuna, though eventually the main centres of brahmin culture, could not have had any significant settlements till iron became relatively plentiful - not before the 8th century B. C. The first "Aryan" settlements were in upper Punjab and along the Himalayan foothills. Ban~ras is perhaps the earliest of the riparian states. R~jgir owed its position to the great metal deposits which lay close and to the south-east. The control of metal sources rather than brahmin organization of vast confederacies explains why Magadha was the first "universal" empire in India. The "masses du Dekhan" did not exist. Though Pai.tha.n was the terminus of the dakkhin, dpatha (southern) trade-route from Kosala, the Deccan plateau was not opened to extensive agricultural settlement till late in the 6th century B.C., and could earlier have provided neither hunting nor pasture comparable to the best northern territory. The coastal strip with its terrific rainfall and heavy forest was developed after A~oka. The pre-Aryan invasions meant at most a relatively thin
35 The data will be found in New Scientist No. 251, Sept. 7, 1961 ; p. 566.

192

D . D . KOSAMBI

scattering of stone-age people, except for the Indus valley. Even here, the light plough or harrow and flood irrigation must have been the norm; the absence of good ploughs and canal irrigation may be deduced from the low density of ancient urban ruins in Sind and the lower Punjab as compared to iraq. Any preponderance of Aryans in number could only have been due to their ability to colonize lands undeveloped before their time, particularly the wooded foothills of the upper Punjab and the Gangetic basin; not that they came to India in great numbers, but that they bred faster and had a higher expectation of life because of the improved and more regular food supply. Aryanization thereafter means primarily the progress of plough agriculture in fixed land holdings - with a new social organization to correspond. The only people that adopted this without the Aryan idiom are Dravidians, not Austro-asiatics. So far as I know, neither the primitive Australians nor those aborigines whose languages (e.g. Munda, Khmer &c) serve as source-material for the Austro-asiatic theory produced any striking innovation in food production. Whatever they know of serious agriculture, metal work, pottery and handicrafts (except weaving baskets and fishing-nets) seems to have been learned after the "Aryan invasion", so that they still remain nearer to the foodgathering stage than any other people in the East. The Udumbara tree (Ficus glomerata) is native to India. Its sanctity, use of its wood for royal consecration thrones, and its edible fruit indicate that it was a totem tree. In fact, there is a historical Udumbara tribe on whose coins 3~ a tree normally appears, presumably the udumbara. There still exist low-caste Udumbaras in Gujar~tt and a few Udumbara brahmins as well. The great Sanskrit poet and dramatist Bhavabhflti was such an Udumbara brahmin? 7 This does not mean organization by the brahmins of a "vast confederation" but that brahmins were adopted into the tribes, or joined the tribal priesthood. This process continued down to the last century ss and is in fact the principal method whereby successive developing groups of dt.avika savages were enrolled as endogamous castes into as For Udumbara coins, J. Allah's Catalogue of Coins in the British Museum, Ancient lndia (London, 1936) pp. 122-123. The legend is Odumbara and the region the Beas valley of the Punjab. s7 The s~ttradhara in the prologue to the drama M~latimftdhava says that the poet belonged to a group of brahmins settled at Padmapuram in the south (dakr TaittirfyflO Ka~yapft~; Udumbara-namana.h. as The most recent example known to me is of the Tigal.as, whose tribal fertility rite was given respectable ancestry by a brahmin during the second half of the last century, and is now the most impressive popular festival at Bangalore.

COMBINED METHODSIN INDOLOGY

193

general Indian society - the formal aspect of "Aryanization", even in Dravidian regions. 5. Sakadani. The classification of ancient Indian peoples on a slender linguistic basis into Aryan and non-Aryan or pre-Aryan groups often excludes the possibility of consistent statements about customs, manner of life, or ethnic affinities. The Brahui "island" in the north is explained on the basis of a pre-historic Dravidian population all over the country. Actually, there is no reason to treat it as other than a casual survival of unabsorbed trading settlers from the south in historic times. Tolstov's excavations at Khorezm show unmistakable south Indian types in stucco relief depicting soldiers on garrison duty for the earlier Ku.sfi .has in Central Asia; the find is supported by anthropometry of the skulls dug up at the site. Alberfmi ~9 refers to Kanarese soldiers in the armies of MahmOd of Ghazni. Adventurers from the Dravidian section of the Peninsula had set up considerable factions at various courts, by the 1 lth century, even in Bengal. a~ Unless the existence of Brahui can be proved, say in the I I I r d millennium B.C. in about the same place as to-day, the linguistic explanation lacks force. The thesis becomes still less convincing when the Burushaski "island" on the K a r a k o r u m is taken into account. The assumption that the non-Aryan and non-Dravidian languages of India, all primitive tribal idioms, can be grouped together as having a common or similar "structure", whatever that term may mean, is doubtful. Przyluski (JRAS, 1929. 273-279) derived Prakrit sdtakan, i from kon " s o n " (Munda) and sadom "horse" (Santali, Mundari, &c), as "son of the horse". He notes the horse emblem on certain Sfttavfihana coins, then the Vi.s.nuite-gaiva conflict and the flowering of Prakrit under a ~fttakar.ni Hfda. The conclusion is : "Quand on voudra mesurer la part des influences anaryennes dans le d6veloppement de la litt6rature prakrite, on ne devra pas perdre de vue que l'onomastique des Andhras contient un important 616ment austro-asiatique". 89 E. Sachau (trans.), Albiranf's India 2 vol. (London 1910); vol. 1, p. 173. For dark-skinned guardsmen at Toprak-kala (Tolstov's excavations) in the 3rd century A.D., see A. Mongait, Archaeology in the USSR (Moscow, 1959) p. 272. The wide extent of the Kushan empire not only made it possible to bring in soldiers from great distance, but even attracted mercenaries from beyond the imperial frontiers. 40 The Senas who superseded the Palas in Bengal were apparently of southern origin; Gftflgeyadevaof Tirabhukti seems to have had Kanarese ancestors; some P~la queens and princes are named in Kanarese style, and the final stanzas of the drama Ca.n.da-kau~ika imply that the wiles of the Nandas were practised at the GurjaraPratih~ra court by Kanarese nobles (cf. the introduction to the Subhas.itaratnako.sa of Vidy~kara, HOS vol. 41, Cambridge, Mass. 1957).

194

D.D.

KOSAMBI

This slipping off into a groove spoils an otherwise valuable study. There is no question that the Sfttavfthanas rose from low tribal origins. ~: Their region, as has been explained, had no agriculture to speak of before the 6th century B.C., hence could not have supported anything beyond small tribes with petty chieftains; certainly not an " A r y a n " king. The horse introduced by Spaniards in America ran wild, bred in large numbers, and was then used by Amerinds of the prairies, who thereby became more efficient in killing the bison. The Aryan horse would similarly have reached some aborigines in the Deccan, or been acquired from northern caravan merchants by way of trade. The tribe or family groups who first used horses would gain superiority in warfare and the hunt. Sgtaka.ni would be equivalent with "horse totem", which agrees with Przyluski's findings; but the Austro-asiatics are superfluous, inasmuch as the totem is found with the horse all around the old world, from the White Horse of the Saxons to the clan name M a among the Chinese. The development into Sfttakar .hi and Sfitavfthana is of peculiar interest. The name is apparently a direct Sanskritization of sdtakan, i by late writers in possession of extensive and beautiful Prakrit literature, but ignorant of the actual dynasty whose tribal origins had vanished into dim antiquity. The Kalki (anu-bhdgavata) Purd.na42 reports a Saptivdhana king named gagidhvaja, who gave his daughter to Kalki. That Kalki was a minor historical character later promoted to a messianic future avatdra is clear from all extent narratives; he was the son of a brahmin and a woman of the low Mgtaflga caste (our ibhyas again) and his symbol is the white stallion. Sapti is good vedic Sanskrit for horse, with special reference to the sacred horses of the sun-god's chariot. Both sapti and saptan "seven" could be prakritized as sdta; the natural confusion may account for the seven horses of Sflrya, who is called saptasapti and so depicted in m a n y icons. The vdhana "vehicle" of an Indian deity is generally shown as his mount, but is obviously a totemic maniThe low tribal origin of the Sfttav~hanas is preserved in Jain tradition, e.g. Rajagekhara-sftri's Prabandhakoga(ed. Jina Vijaya, ~ntiniketan 1935; Singhi Jaln Series 6), story 15. The original S~tav~hana was born of a brahmin widow ravished and impregnated by the ndga (cobra-demon) of a pool in the Godf~variriver; Paithan was then a hamlet, and the widow's two brothers lived there by some sort of food-gathering. Tftran~tha (in A. Schiefner's translation) similarly reports a naga father for the first P~la king. With the Mbh heroes, of course, we have the immaculate conception in the manner of Trobriand islanders, which means that the father was traditionally unknown, fatherhood then being of no importance; the Mbh tradition must basically have been pre-patriarchal, hence pre-Aryan. 4, In the printed edition (without frontispiece, Bengali form of devandgarf type) 8.1 ; the SaptivS_hanais given as king of Bhall~tanagara.
4:

COMBINED METHODSIN INDOLOGY

195

festation of the god or goddess. Thus Brahmri is the swan. Clear evidence of pre-historic and pre-Siva worship of the humped bull a3 has been uncovered by archaeologists. The large animal which normally occupies the greater part of an Indus seal is presumably a clan emblem, just as the Athenian Boutadai had their shields marked with a bull's head. There is a direct line of descent from the prehistoric ice-age artist's pebble "sketch-sheet" and the stamp seals and cylinder seals used to protect merchandise from Mesopotamia to the Indus. Saptikartza "horse-ear" sounds like a "split totem" which sometimes develops when a primitive exogamous clan splits into two or more units. The clan name Ghot.aka-mukha "horse-face" occurs in the gotra lists and the K r i m a s u t r a y while Gho.tamukha is reported in Arth. 5.6 as a former master of political science. Earlier, the legend of ~una.h-~epa and his brothers, each of whose names means "dog's tail" and famous gotra names like Saunaka (from ~van "dog", ~unaka " p u p p y " ) carry one in the same direction. There is actually a Sanskrit word for "split clan", namely gotrdvayava (Prig. 4.1.79). In Prig. 4.1.173 the Udumbaras and others are (according to commentators) avayava components of the Srilvas; this is treated as a confederacy by Przyluski, but the two possibilities are not mutually exclusive. The etymology of gotra "cowpen" and the comment on Prig. 4.3.127 implies that at some stage, the local gotra group had a distinguishing m a r k for its men and brand for cattle presumably owned in common. gfilva is given as tree with edible fruit by some commentators on Prig. 4.3.166; a large number of brahmin gotra names 4~ are edible tree- or animal-totems as among so many savages and for that matter among Latin gentes. We shall consider here only six examples of Sanskrit names ending in kar.na, none in the same category as manda-karn.a "hard of hearing". In the gan.a Sivridi (Prin.. 4.1.112) are found (in the K6~ikd also) the clan names t.r.nakar.na (var. tfma-), may~rakarn, a, mas~trakarn, a, kharj~rakar.na; respectively "grass-ear", "peacock-ear", "lentil-ear," "dateear". These exclude the split totem; nor can they be used to describe shape ~8 The latest such excavations known to me were by F. R. and Mrs. B. Allchin at Piklihg]; their fial report has not yet come to hand. 44 Gho~akamukha is reported in K~rnasutra 1. 114 as the authority for the third section of that work. Hayagriva and Hayavadana may be adjectives, and Haihaya may or may not be connected with the horse, in spite of the termination. 45 The best available gotra fists are in J. Brough; The early brahmanical System of Gotra and Pravara (Cambridge, 1953); actual gotras found in Mah~r~.st.ra among the Degastha brahmins have been collected by V. T. ~eSe in his Gotr~vali (in Mar~hi; Ygjfiavalkya .~Arama, Poona 1951).

196

D . D . KOSAMBI

or colour of a human ear. The analogy with saptikarn, a is clear, and one may point to a daunakarn,i "son of dog-ear" in the gotra lists. Still better known is Yatf&arn.a "bat-ear". In each of these cases, the termination - karn.a signifies "descent from" rather than a split totem. Finally, the demon Kumbhakarn. a might have had ears like pot-handles (e.g. the Scottish "lug"). But the kumbha is often the homologue of the uterus and symbolizes a mother-goddess. This would explain the otherwise stupid account of the hundred Kaurava sons and one daughter o f Gfmdhftri being born through the intermediacy of ghrta-kumbha gheejars; that many of these sons were patron yak.sa cacodemons of northern towns 46 is known. Vasi.st.ha and Agastya had similar origin, being born from womb-jars, and the dcdrya Dro.na's name as well as birth-story throws him in the same category. Draun.i and Drau.ndyana are again listed as gotras. Dron.a's son A~vatthfiman bore in his forehead (from his very birth) a precious jewel - the symbol of a ndga. So, the Sanskrit termination -karn.a can signify "son of" as in Mun..dftri, and may be associated with pre-Aryan elements. That a man has a good Aryan name does not mean that he had an Aryan father, nor even that he had a father at all.

6. Parallel development. It might seem at this point that I merely replace Austro-asiatic by ndga or some such change of name. The matter lies much deeper, being the gradual and progressive absorption of many distinct dt.avika tribes into general indian society which had had its own course of food-producing development since 3000 B.C. The influence of food-producing neighbours, infiltration by caravan merchants, Buddhist, Jain and other monks, brahmin priests and an occasional adventurer of some military capacity would generally introduce food-production and a class structure. From that stage, the course of assimilation depended upon the relative wealth and armed strength of the environment. The important point is that there was always a reciprocal influence. It seems to me that forgotten tribes show their existence in the onomasticon of peasant deities, particularly the mother-goddesses; Sirkfti, Tuk~i, Bolh~i, Mefigfti, Sofigzft'i, Kumbha.lj~ (and of course the pre-gakyan Lumbini)
46 SylvainL6vi, Le Catalogue g6ographique des yak.sas darts la Mah~tm~ty~ri;Journal Asiatique 5. 1915 (i). 19-138; line 23 of the Sanskrit text, Duryodhanas ca Srughne.su; but the list is composite, probably from many different sources: 1.60-siddhaydtras tatha Srughne. For Bharukaccha, Bharuka in 1.17, Asafiga in 1.43; for R~tjag.rha, Vajrap~.ni in 1.3, Bakula in 1.6., Kumbhira in 1.101. Not that there need be only one yak.sa per city, but the principal guardian could be only one - here a different one for each particular tradition among the worshippers.

COMBINED METHODS IN INDOLOGY

197

seem to have no acceptable derivation. The folk etymologies are demonstrably eponymous, sometimes as crude as the word-derivations in the Br~thma.nas and Upani.sads. But there is nothing to show that any of these were Austro-asiatic nor that they all belonged to one pre-Dravidian or pre-Aryan group. Brahmin tradition lumped all kinds of aborigines together under the generic title ndga (cobra or more rarely elephant), presumably as snake-worshippers. The ndga cobra becomes a garland for giva, bed and canopy for Vi.sn. u, the patron demon for many Buddhist vihdras and a few cities. The mother-goddesses are, whenever the number and wealth of their worshippers warrants it, identified with Durgh, Lak.smi, or the like, "married" to the corresponding god and worshipped in suitably endowed temples. This brahminization reflects the underlying change from food-gathering in independent tribal units to food-production in a society that preserved endogamy and a (hierarchical) commensal tabu as features of its caste system. This preservation is due primarily to the fact that food-gathering remained a powerful supplement to agriculture till the forests disappeared, while clothing and shelter are not physically indispensable over most of India. It should be noted that Indian monastic tradition also has deep roots in the food-gathering tradition. The danger of treating "Aryan" as a homogeneous unit over any considerable extent of time or space, or even in any large literary source formed over many centuries, may easily be demonstrated. The Madra tribe in the Mahfibh~trata was settled in the north-west, along with the allied ghlva, Udumbara, B~thlika and Gfindh~tra. Both P~t.l:tini and Patafijali came from or near this territory. The more learned Upanisadic philosophers (B.rhad.3.3.7 and 3.7.1) claimed to have wandered among the Madras to study the yaj~a fire-ritual, the very core of the sacred vedas. The local host is named as Pataficala Khpya. Jfitaka tradition supports this independently in placing Taxila as the main center of (vedic, Sanskrit and medical) education to which Gangetic princes and brahmins travelled by the great northern trade route, the uttardpatha. For that matter, the Upani.sads (Chhnd. 5.3~5.11; B.rhad. 2.1.6.2.) show brahmins at K ~ i and Paficala learning high philosophy from k.satriyas; a perfectly genuine though unbrahminical tradition continued in history by great Magadhan k.satriya teachers like the g~tkyan Buddha and the Licchavi Mah~tvira. Nevertheless, Kar.na as the rttler of Afigh in the east exchanges biting discourtesies with king galya of Madra-land, though the latter has agreed to act as Karn.a's charioteer in the imminent desperate and hopeless contest. The reproaches against the Madras and their neighbours are

198

D. D. KOSAMBI

that: Women mixed freely with men, without restraint or modesty. All drank and ate meat. The ladies would cast off their garments to dance when intoxicated ... Still more shocking was slackness in observance of caste distinctions (8.30). "There a B~hlika who has been a brahmin becomes a k.satriya, a vai~ya or ~Odra, or even a barber. From a barber he again becomes a brahmin. Having been a twice-born (dvija), he there becomes a ddsa again... In the same family one (male) may be a brahmin while the rest are common workmen". It does not seem to have struck the brahmin redactors of the Mbh, nor for that matter Salya himself, that this kind of abuse sat ill in the mouth of Kar.na. Though ranked as a pre-eminent k.satriya, Karn.a had no legal father, had been exposed by his unwed mother to hide her shame, rescued and brought up as his own son by a lowly professional chariot-driver. The censure only proves that the Madras and their allies retained the older Aryan custom whereby no man was degraded by his profession, while ritual had to be performed by some member of the family or clan. (Parenthetically, this last rule alone can explain the presence of so many tribal names in the brahmin gotra list, whether the brahmins were originally strangers adopted into the tribe or members of the tribe who specialized in pontifical functions). The quotation agrees very well with sutta 140 of the Majjhima-nik~ya. The P~li discourse reminds the brahmin Assal~tyana through the mouth of the Buddha that in Yona, Kamboja, and other regions beyond the (north-west) frontier, there were only two castes: Arya (=free) and ddsa (=slave); moreover, a person who had been an Arya could become a ddsa and conversely. That is, the MadraB~thlika-Gandh~ra-Kamboja lands had developed a form of chattel slavery nearer to the classical Graeco-Roman model than to the complex and rigid caste system evolved in the Gangetic plain. As explained, the latter was better suited for the peaceful absorption of savage tribes in the warmer and wetter parts of India, under the conditions that prevailed before mechanised production became the norm. This cumulative difference had become significant by the end of the 4th century B.C. Earlier in the great epic, a Madra princess famous for her beauty had literally been purchased by Bhi.sma as legal wife for his nephew P~m..du, with no more ado than over a basket of vegetables: Pd.n.dor arthe parikr[td dhanena rnahatd tadd (Mbh. 1.105.5). This passage proved so embarrassing to later brahmin orthodoxy that several versions of the Mbh insert discordant interpolations to explain it away. The sm.rtis forbid brideprice for the upper castes (Ms. 3.51-3) as amounting to the sale of a daughter; therefore, in the high ar.sa form of marriage, the gift even

COMBINED METHODSIN INDOLOGY

199

of a pair or two of cattle to the bride's father was forbidden (Ms. 3.53). Nowhere is the wedding of Mftdri declared Asura as it would be by Ms. 3.31; it might be added that the custom is permissible in some 80 or more of the Mah~trft.st.rian population; brahmins do not hesitate to officiate (for a consideration) at such weddings. The change from .Rgvedic to Yajurvedic Aryans corresponds rather well to that between the ruder Germani of Tacitus and Caesar's Gauls of the later La T6ne iron age culture. This is another example of parallel development, not a suggestion that the Druids were really brahmins or that Caesar must be later than Tacitus! When we look for totemic origins in the gotra lists, there is no implication that the brahmins concerned were comparable to medicine men of Austro-asiatic savages. Nevertheless, brahmin penetration of the priesthoods of comparatively savage groups is demonstrable or deducible from the earliest "Aryan" period down to the last century. The Manusmrti interdict at a feast for the manes upon any brahmin who sacrificed for tribal organizations gan.dndm. caiva ydjaka.h (Ms. 3.164) would otherwise have been quite superfluous. How explain the ~aigrava gotra (attested by a Mathurft inscription ~7 though absent from surviving gotra lists) among brahmins except by association with the ~igru tribe of the .Rgvedic (RV. 7.18) Ten Kings' War? Is not the tabu upon the horse-radish digru ("Moringa pterygosperma") as food for ascetics (Ms. 6.14) of such tribal-totemic origin? The iguana is specially excepted (Ms. 5.18) from the tabu on the flesh of five-nailed creatures, but eaten today only by the lowest castes; what of godhddana "iguana eater" as a gens in the gan.a Kftgygdi (on Prin.. 4.2.116)? The hungry brahmin wanderer Baka Dfilbhya (or Glgva Maitreya) spies in Chfmdogya Up. 1.12 upon an assembly of dogs, led by a white dog (dvd &eta.h) as they dance hand in hand to perform an udgftha chant for food. This can only mean a fertility rite of a dog-totem clan; I have witnessed similar chants and dances among the lowest Indian tribal castes. A Kukuraka ("dog") tribe is listed among the formidable military tribes in Arth. 11 ; a cut above the at.avikas but dangerous to royal power. The historical name Kokerah for the region about Ranchi in Bihar may be due to the Kukurakas. We have already noted the brahmin daunaka gens. In the same way, modern linguists talk of a Ko.1 language or group of 47 H. Ltiders noted in reading the Mathurfi inscription (Epigraphia Indica 9.247-8) that the brahmin of the Segrava gotra there named was treasurer of the ~aka k.satrapa king .So.d~sa; the title ga.mjavara, of which this seems to be the earliest mention, is a loan word from the Persian ganjwdr. Liiders further comments that the legendary preceptor of acdrya Moggaliputta Tissa was a Siggava.

200

D.D. KOSAMBI

languages. A Ko.liya tribe is clearly referred to in the Jgtakas as having the Ko.1 tree Zizyphus juluba as a totem; the Sanskrit name badara for the same jujube tree leads to Bftdar~ya.na, whom no one relates to the Koliyans. In Marfi.thi, Kol.i (like ndga further north) means the originally heterogeneous marginal tribe-castes that took late in history to agriculture and were often pressganged for porterage in army service. The same word also means spider and fisherman, presumably because the fisherman makes and uses a net to catch his prey as a spider his web. Here the derivation is not totemic but occupational; heavy deposits of microliths at certain favoured spots on the river bank surely indicate pre-historic fishing camps in MahftrS..stra. Men of the Ko]i caste still catch fish and keep up age-old cults at some of these places, as at Cfts-Kamfin. The ~fikyans seem closely related to the ddka tree (Shorea robusta) and there existed two subgroups among them known as reed-ggkyas and grass-~fikyas, the last being reminiscent of t.rn.akarn, a. Pippalgda as a gotra has a modern nonbrahmin counterpart among the Pimp.16s (now a surname, once a clan) who, at their village Pimp.loli, still observe characteristic tabus such as not eating off plates made ofpim,pal (Ficus religiosa) leaves. This should place the Udumbaras in proper perspective. There still exist tiny remnants of a gava[~ tribal caste, who live solely by pasturing cattle. To most city dwellers gaval.f means only "milk-man" whatever his caste. Remote villages report strong traditions which show that the now extinct gaval.fs were relatively more numerous at one time and relatively more important in the rural economy. This sounds like an Aryan invasion, but I have been unable to find any indication of their possessing horse-chariots, the heady soma drink, the overdeveloped fire ritual or the powerful aggressive tendencies of vedic Aryans. Archaeologically, their successive waves appear in the western Deccan to be responsible for megaliths, rock-engravings of a peculiar type, upland terraces not meant for the plough, and certain remarkable mortarless structures (v&dagg) of undressed stone that are traditionally cattle enclosures though never used as such. The terraces and vdd. ag~ are sometimes ascribed to the mythical Age of Truth (satya yuga) by older peasants. Occasionally, the pastoral cults survive in the name of a comparatively rare patron god of cattle: gava[~fi-bdbd. Still rarer is the use of the term to describe a village. One such is Gava!yfici U.n.davadi not far from Bgtfftmati, with a companion village Corftci U.n.dava.dL The village Cor~ci _&.landi has a tradition that the qualification "thief's" was originally genitive plural: cordm,d " o f the brigands". The origin of this latter village can be traced back to long before the 8th century A.D. The added cora both at A.landi

COMBINED METHODS IN INDOLOGY

201

and U.n.dava.di merely denotes a settlement of tribal origin which long retained habits of brigandage, taking to plough culture much later than neighbouring villages. This would be impossible to restore without field work, merely from the etymology of gava[f and cora; in the latter case, distant villagers invent some repentant thief who originally settled the village of A.landi. The primitive goddess Bolhfti is reported by her senior worshippers, the Vaji ("horse") clan at Pusa.ne, to have been taken by coras to her present location, which represents tribal cult migration quite accurately. A modern observer could report (New Yorker, April 18, 1959, p. 119) that in the neighbourhood of Pawa in northeastern Congo: "The pygmy women used a kind of sing-song in their speech.., and there were experts who believed that this was the vestige of an ancient pygmy language; nowadays the pygmies had no identifiable language of their own, merely speaking that of whatever settled tribe they lived near .... They had a natural balance of trade - the sort of mutual dependency that naturalists call symbiosis. The pygmies killed game and gave some of it to the villagers, whose normal diet lacked proteins, and in return got the products of agriculture - mainly bananas - which, as nomads they did not grow themselves. Nowadays... the pygmies are accustomed to a steady supply of bananas and this keeps them from disappearing into the forest for very long. The men may hunt for days on end, but meanwhile the women will go back to the villages to fetch bananas and this ties them all down to some degree." No better illustration could be found of the development of primitive languages in relation to food gathering and food production. Now add the following important remarks by T. Burrow (Trans. 19, Bull. Ramakrishna Mission Inst. of Culture, Feb. 1958): "The number of loan-words in Sanskrit, which cannot be explained as either Dravidian or Munda, will remain considerable. It may very well turn out that the number of such words which cannot be so explained will outnumber those which can be. This is the impression one gets, for instance, from the field of plant-names, since so far only a minority of this section of the non-Aryan words has been explained from these two linguistic families. If we take, for instance, the name of the jujube (Zizyphusjujuba), we find four synonyms, all obviously non-Aryan words, namely kuvala or kola, karkandhu, badara and ghon.t.d; and none of these has been explained out of either Dravidian or Munda. Evidence such as this leads to the conclusion that there must have been several non-Aryan languages or families of languages which exercised an influence on the vocabulary of Indo-Aryan". Inasmuch as the total number of words in

202

D.D. KOSAMBI

use has grown with social production, it may be better to concentrate upon parallel development rather than invent fictitious origins. Language is surely a means of exchanging ideas, which cannot precede the exchange of surplus. This implies that any language common to more than a handful of people must have been preceded by commodity production and exchange on a corresponding scale. But it is known that, in the most primitive societies, such exchange is not simple pubic barter with a basic standard of equivalents modified by haggling or by the laws of supply and demand. On the contrary, the exchange appears at its most primitive level in the form of gifts that cannot be refused and must ultimately though not immediately be compensated by a reciprocal gift from the recipient. Moreover, these gifts are only to be made between fixed persons48 of different tribes in a special relationship, "trade friends". Within the tribe, such gifts are obligatory, dependent upon the status of the giver, with no idea of compensation - a form of distribution of the surplus. It seems plausible that at a still earlier period, the tribe was fused out of individual totems on the same basis, with exchange of human beings in some form of exogamous "marriage" as a concomitant of the transfer of food (often the special totem product) or techniques. If so, the development of language cannot be separated from the succession of pre-historical stages through which a given society has passed. The position stated does not approach the formalism of Marr's ~aphetic Theory which derived all Caucasian languages and perhaps all languages from the four mystic syllables yon, bet, sal and ro~. It differs also from the Durkheim-Levy-Bruh149 type of sociology which takes "pre-logical" mentality as a fixed characteristic of certain ethnic groups, not as the concomitant of the various stages of development through which the particular group reached its actual level of social production. One may leave out of discussion the higher mentality which takes slums, world wars, massive colonial suppression and nuclear bombs as logical assets of civilization; but two questions remain. Did not the superior "logical" people once pass through the same "pre-logical" stage, say when their ancestors could make only the simplest tools of stone? Secondly, what caused the change from the pre-logical to the logical mentality? One possible answer has been suggested in this note. 48 See particularly, B. Malinowski, Crime and Custom in Savage Society (London) 1940, pp. 22-25; also, Margaret Mead: Sex and Temperament in Three Primitive Societies (MentorBooks, New York 1950): pp. 19-20 and 31 for the Arapesh. 49 L. Levy-Bruhl,Les Fonctions mentales dans les socidtds inf~rieures, translatedinto English as: How Natives Think (New York, 1925).